44 REPORT — 1839. 



at a distance of about thirty years from the period when attempts to 

 ascertain such weights with accuracy were first made, to have placed 

 before us, not merely the mean or selected result of the fundamental 

 experiments, but also their highest and lowest results. The variations 

 are much greater than chemists in general are aware of. 



To attempt giving a view of the variations in the atomic weights of 

 each elementary substance would occupy far too much time for an oc- 

 casion like the present. All that I propose is, by adverting to certain 

 compounds of lead, in illustration of the atomic weight of that metal 

 and of the atomic weights of sulphur, azote, and carbon, to afford some 

 idea of the limits within which the atomic weights of these four ele- 

 ments have been ascertained. 



In proceeding to indicate such limits, I need not, in this presence, 

 express how much the selection of safe data requires the exercise of 

 circumspection. The results of only such experiments as have been 

 performed with the utmost care should be employed. By including 

 the results of experiments otherwise conducted, we increase the limits 

 of the recognisable variations, which is the opposite of desirable, with- 

 out thereby increasing the certainty of the mean or selected result. 



The Protoxide of Lead. 

 The proportions of oxygen and lead in this compound have been 

 repeatedly examined by Berzelius. The mode of analysis ultimately 

 preferred by him, consisted in reducing the oxide by pure hydrogen 

 gas, at an elevated temperature, and weighing the residual metal. The 

 loss Avas accounted oxygen. In his last and most careful experiments, 

 he obtained, as the result of six experiments, for every 100* of oxygen, 

 the following quantities of lead : 



Greatest 1295-70^ 



Mean 1294-24 > Weighed in air. 



Least 1293-08 J 



But inasmuch as, weight for weight, the protoxide of lead is more 

 bulky than lead itself, the oxide would, when weighed in air, be pro- 

 portionably more buoyed- up than the metal. Thus if 1394- grains of 

 lead were balanced in air against 1394- grains of its protoxide, the same 

 1394- grains of lead would, when transferred to a vacuum, counterpoise 

 1394-033 grains of the same protoxide. Hence, in the foregoing results, 

 the oxygen, instead of 100*, should be reckoned 100-033. By reducing 

 this 100-033 to 100-, and by reducing the lead in the same proportion, 

 we obtain of lead to every 100- of oxygen. 



Greatest 1295-271 



Mean 1293-82 > Weighed in a vacuum. 



Least 1292-65 J 



Berzelius has, more than once, observed that the circumstance of the 

 atomic weight of lead being about thirteen times the atomic weight of 

 oxygen, causes the slightest error in experiment to be very much mul- 

 tiplied in calculation. Thus the difference between the highest and 



